Production preparation method of a special concrete

Abstract

The present application relates to a production preparation method for special concrete. The method includes: collecting an image of the stirring state of the first cement mixture by a camera, and using image recognition and analysis technology based on computer vision to perform feature analysis and multi-scale perception enhancement on the stirring state image, so as to automatically judge whether the stirring state meets the predetermined requirements based on the fusion expression between the stirring state enhancement features of the first scale and the second scale, and determine whether to add stone materials and reinforcing fibers to the first cement mixture in batches. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of the concrete can be ensured, and at the same time, the addition timing of the materials can be accurately controlled to avoid premature or late addition, thereby improving the intelligent level of the mixture stirring.

Description

Technical Field

[0001] This application relates to the field of construction technology, and specifically, to a production preparation method of special concrete. Background Art

[0002] With the increasing intensity of mineral resource development, the tailings discharge continues to rise. Although traditional harmless stacking can control the spread of pollution, it has not fundamentally solved the problem of accumulation. At the same time, a large amount of fly ash generated by the thermal power industry also poses an environmental challenge. Although the resource utilization of tailings and fly ash is extremely urgent, due to their complex composition and large treatment volume, there is currently a lack of efficient, environmentally friendly, and economically practical large-scale comprehensive utilization technologies. And the existing concrete production equipment has insufficient capacity in material grinding and screening, resulting in limited concrete quality and difficulty in meeting diverse construction requirements.

[0003] In response to the above technical problems, Chinese Patent CN113816656B proposes a special concrete based on waste resource utilization and its preparation process. By grinding tailings containing quartz, it promotes the transformation of quartz molecules into active disordered glassy states, enhances the gelling reaction with cement, and improves the hardness and wear resistance of concrete. At the same time, an appropriate amount of fly ash is incorporated to optimize the comprehensive performance of special concrete, thus solving the problems of insufficient traditional resource utilization and heavy environmental burden, realizing the resource utilization of tailings and fly ash, and reducing production costs.

[0004] During the preparation process of the above special concrete, it is necessary to add stones and reinforcing fibers to the cement mixture and stir to enhance the adhesion between the mixtures and improve the overall performance of the concrete. However, in the process of monitoring the stirring and mixing state to determine when to add stones and reinforcing fibers, the above patent may rely on the operator's observation and experience and use a basic timer for evaluation and judgment. However, since each operator's understanding and judgment of the stirring uniformity and material mixing degree may vary, this will directly affect the uniformity and strength of the concrete. In addition, using a timer to control the addition timing of materials provides a certain degree of standardization, but this method ignores the changes in the actual stirring state and cannot flexibly adjust the addition time point, which may lead to uneven material mixing or over-stirring, thereby affecting the final performance of the concrete.

[0005] Therefore, an optimized production preparation method of special concrete is desired. Summary of the Invention

[0006] This Summary of the Invention section is provided to introduce concepts in a concise form that will be described in detail in the subsequent Detailed Description section. This Summary of the Invention section is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] In a first aspect, the present application provides a method for producing and preparing special concrete, and the method includes:

[0008] Grind and pulverize tailings and fly ash to obtain ground tailings and ground fly ash;

[0009] Mix the ground tailings and the ground fly ash through stirring to obtain a raw mixture;

[0010] Add cement to the raw mixture and stir to obtain a first cement mixture;

[0011] Add stone materials and reinforcing fibers to the first cement mixture and stir to obtain a second cement mixture;

[0012] After uniformly mixing a water reducing agent and water, add them to the second cement mixture and stir to obtain special concrete;

[0013] Wherein, adding stone materials and reinforcing fibers to the first cement mixture and stirring to obtain a second cement mixture includes:

[0014] Obtain a stirring state image of the first cement mixture collected by a camera;

[0015] Pass the stirring state image through a stirring state feature extractor to obtain a stirring state feature map;

[0016] Perform multi-scale perception enhanced expression on the stirring state feature map to obtain a first-scale stirring state perception enhanced feature map and a second-scale stirring state perception enhanced feature map;

[0017] Perform global average pooling processing on each feature matrix along the channel dimension in the first-scale stirring state perception enhanced feature map to obtain a first-scale stirring state perception enhanced feature vector;

[0018] Pass the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map through a gated multi-scale stirring state fusion module based on a transformer structure to obtain a stirring state multi-scale fusion expression feature map;

[0019] Based on the stirring state multi-scale fusion expression feature map, obtain an identification result, and based on the identification result, judge whether to add stone materials and reinforcing fibers to the first cement mixture in batches.

[0020] Optionally, passing the stirring state image through a stirring state feature extractor to obtain a stirring state feature map includes: passing the stirring state image through a stirring state feature extractor based on an atrous convolutional neural network model to obtain the stirring state feature map.

[0021] Optionally, perform multi-scale perception enhancement on the stirring state feature map to obtain a first-scale stirring state perception enhancement feature map and a second-scale stirring state perception enhancement feature map, including: performing convolution processing on the stirring state feature map based on a multi-layer convolutional layer to obtain a shallow stirring state feature map and a deep stirring state feature map; performing convolutional encoding on the deep stirring state feature map to obtain the first-scale stirring state perception enhancement feature map; performing upsampling processing on the first-scale stirring state perception enhancement feature map to obtain an upsampled first-scale stirring state perception enhancement feature map; and performing multi-scale feature perception processing on the shallow stirring state feature map to obtain the second-scale stirring state perception enhancement feature map.

[0022] Optionally, perform multi-scale feature perception processing on the shallow stirring state feature map to obtain the second-scale stirring state perception enhancement feature map, including: inputting the shallow stirring state feature map into a multi-scale perception network to obtain a shallow multi-scale perception stirring state feature map; fusing the features of the shallow multi-scale perception stirring state feature map and the upsampled first-scale stirring state perception enhancement feature map, and performing convolutional encoding on the obtained stirring state fusion feature map to obtain the second-scale stirring state perception enhancement feature map.

[0023] Optionally, inputting the shallow stirring state feature map into a multi-scale perception network to obtain a shallow multi-scale perception stirring state feature map, including: copying the shallow stirring state feature map to obtain a shallow stirring state copy feature map; first passing the shallow stirring state feature map through a convolutional layer with a convolution kernel of 1×1, and then sequentially performing activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a first-scale shallow stirring state feature map; first passing the first-scale shallow stirring state feature map through a convolutional layer with a convolution kernel of 3×3, and then sequentially performing the activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a second-scale shallow stirring state feature map; first passing the second-scale shallow stirring state feature map through a convolutional layer with a convolution kernel of 5×5, and then sequentially performing the activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a third-scale shallow stirring state feature map; and cascading the shallow stirring state copy feature map, the first-scale shallow stirring state feature map, the second-scale shallow stirring state feature map, and the third-scale shallow stirring state feature map through a feature fusion layer to obtain the shallow multi-scale perception stirring state feature map.

[0024] Optionally, the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map are passed through a gated multi-scale stirring state fusion module based on the transformer structure to obtain a stirring state multi-scale fusion expression feature map, including: performing layer normalization and convolutional encoding on the second-scale stirring state perception enhanced feature map to obtain a second-scale stirring state perception enhanced spatial deepening feature map; copying the second-scale stirring state perception enhanced spatial deepening feature map to obtain a backup second-scale stirring state perception enhanced spatial deepening feature map; respectively reshaping the feature shapes of the backup second-scale stirring state perception enhanced spatial deepening feature map and the second-scale stirring state perception enhanced spatial deepening feature map to obtain a backup second-scale stirring state perception enhanced spatial deepening feature matrix and a second-scale stirring state perception enhanced spatial deepening feature matrix as an imitation query feature matrix and an imitation key feature matrix; calculating the product between the imitation query feature matrix and the transposed matrix of the imitation key feature matrix to obtain a second-scale stirring state perception enhanced local self-correlation library matrix; inputting the second-scale stirring state perception enhanced local self-correlation library matrix into an attention gated unit based on the GELU function to obtain a second-scale stirring state perception enhanced gated local self-correlation library weight matrix; using the first-scale stirring state perception enhanced feature vector as a value feature vector, calculating the product between the second-scale stirring state perception enhanced gated local self-correlation library weight matrix and the first-scale stirring state perception enhanced feature vector to obtain a stirring state multi-scale cross-modal interaction fusion conditional weight vector; using the stirring state multi-scale cross-modal interaction fusion conditional weight vector as a weight vector, performing channel-wise weighting on the second-scale stirring state perception enhanced feature map to obtain the stirring state multi-scale fusion expression feature map.

[0025] Optionally, performing layer normalization and convolutional encoding on the second-scale stirring state perception enhanced feature map to obtain a second-scale stirring state perception enhanced spatial deepening feature map includes: performing layer normalization on the second-scale stirring state perception enhanced feature map to obtain a layer-normalized second-scale stirring state perception enhanced feature map; performing point convolutional encoding and spatial convolutional encoding on the layer-normalized second-scale stirring state perception enhanced feature map to obtain the second-scale stirring state perception enhanced spatial deepening feature map.

[0026] Optionally, based on the stirring state multi-scale fusion expression feature map, an identification result is obtained, and based on the identification result, it is determined whether to add stone materials and reinforcing fibers to the first cement mixture in batches, including: passing the stirring state multi-scale fusion expression feature map through a stirring state identifier based on a classifier to obtain the identification result, where the identification result is used to indicate whether the stirring state meets a predetermined requirement; in response to the identification result indicating that the stirring state meets the predetermined requirement, adding stone materials and reinforcing fibers to the first cement mixture in batches.

[0027] Optionally, passing the stirring state multi-scale fusion expression feature map through a stirring state identifier based on a classifier to obtain the identification result includes: expanding the stirring state multi-scale fusion expression feature map into a classification feature vector according to a row vector or a column vector; performing fully connected encoding on the classification feature vector using multiple fully connected layers of the stirring state identifier based on the classifier to obtain an encoded classification feature vector; passing the encoded classification feature vector through the Softmax classification function of the stirring state identifier based on the classifier to obtain the identification result.

[0028] By adopting the above technical solution, the stirring state image of the first cement mixture is collected by a camera, and image recognition and analysis technologies based on computer vision are used to perform feature analysis and multi-scale perception enhancement on the stirring state image, so as to automatically determine whether the stirring state meets the predetermined requirement based on the fusion expression between the stirring state enhancement features of the first scale and the second scale, and determine whether to add stone materials and reinforcing fibers to the first cement mixture in batches. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of concrete can be ensured, and at the same time, the addition timing of materials can be accurately controlled to avoid premature or late addition, thereby improving the intelligent level of mixture stirring.

[0029] Other features and advantages of the present application will be described in detail in the subsequent specific implementation part. Brief Description of the Drawings

[0030] Combined with the drawings and referring to the following specific implementation manners, the above and other features, advantages and aspects of the embodiments of the present application will become more obvious. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic, and the original components and elements are not necessarily drawn to scale. In the drawings:

[0031] Figure 1 is a flowchart of a method for producing and preparing a special concrete according to an exemplary embodiment.

[0032] Figure 2 is a block diagram of a system for producing and preparing a special concrete according to an exemplary embodiment.

[0033] Figure 3 is a block diagram of an electronic device shown according to an exemplary embodiment.

[0034] Figure 4 is an application scenario diagram of a production preparation method of a special concrete shown according to an exemplary embodiment. Detailed implementation manners

[0035] Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not used to limit the protection scope of the present application.

[0036] It should be understood that the various steps recited in the method embodiments of the present application can be executed in a different order and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this regard.

[0037] As used herein, the term "including" and its variations are open-ended, that is, "including but not limited to". The term "based on" is "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0038] It should be noted that the concepts such as "first" and "second" mentioned in the present application are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependent relationships.

[0039] It should be noted that the modifications of "one" and "multiple" mentioned in the present application are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".

[0040] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only for illustrative purposes and are not used to limit the scope of these messages or information.

[0041] The following will describe the specific implementation manners of the present application in detail with reference to the accompanying drawings.

[0042] Figure 1It is a flowchart of a production preparation method of a special concrete shown according to an exemplary embodiment. As Figure 1 shown, the method includes:

[0043] Step S101, grinding and pulverizing tailings and fly ash to obtain ground tailings and ground fly ash;

[0044] Step S102, stirring and mixing the ground tailings and the ground fly ash to obtain a raw mixture;

[0045] Step S103, adding cement to the raw mixture and stirring to obtain a first cement mixture;

[0046] Step S104, adding stone materials and reinforcing fibers to the first cement mixture and stirring to obtain a second cement mixture;

[0047] Step S105, after uniformly mixing a water reducing agent and water, adding them to the second cement mixture and stirring to obtain special concrete;

[0048] Among them, Step S104, adding stone materials and reinforcing fibers to the first cement mixture and stirring to obtain a second cement mixture, includes:

[0049] Step S1041, obtaining a stirring state image of the first cement mixture collected by a camera;

[0050] Step S1042, passing the stirring state image through a stirring state feature extractor to obtain a stirring state feature map;

[0051] Step S1043, performing multi-scale perception enhanced expression on the stirring state feature map to obtain a first-scale stirring state perception enhanced feature map and a second-scale stirring state perception enhanced feature map;

[0052] Step S1044, performing global average pooling processing on each feature matrix along the channel dimension in the first-scale stirring state perception enhanced feature map to obtain a first-scale stirring state perception enhanced feature vector;

[0053] Step S1045, passing the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map through a gated multi-scale stirring state fusion module based on a transformer structure to obtain a stirring state multi-scale fusion expression feature map;

[0054] Step S1046, based on the stirring state multi-scale fusion expression feature map, obtaining an identification result, and based on the identification result, judging whether to add stone materials and reinforcing fibers to the first cement mixture in batches.

[0055] It should be understood that the raw materials of the special concrete can be mixed from tailings, fly ash, cement, water reducer, stone, reinforcing fiber and water according to precise proportions. Among them, in step S101, the tailings and fly ash are ground and pulverized to obtain ground tailings and ground fly ash. The tailings and fly ash are ground to a certain fineness to increase their contact area with other materials such as cement, thereby enhancing the reaction activity and bonding force between the materials. The quartz molecules in the tailings are transformed into active disordered glassy state during the grinding process, which helps to improve the hardness and wear resistance of the concrete. In step S102, the ground tailings and the ground fly ash are subjected to a stirring and mixing process to obtain the original mixture. The ground tailings and fly ash are mixed to form the original mixture. The purpose of this step is to ensure the uniform mixing of the two materials and lay a foundation for the subsequent addition of cement and further mixing. In step S103, cement is added to the original mixture and stirred to obtain the first cement mixture. Cement is the main gelling material in concrete, and its addition and uniform mixing are the keys to the hardening and strength formation of concrete. In step S104, stone and reinforcing fiber are added to the first cement mixture and stirred to obtain the second cement mixture. Stone provides the volume and structural support of the concrete, while the reinforcing fiber is used to improve the tensile strength and toughness of the concrete and enhance its overall performance. In step S105, after the water reducer and water are mixed evenly, they are added to the second cement mixture and stirred to obtain the special concrete. After the water reducer and water are mixed evenly and added to the second cement mixture, the special concrete is finally obtained by stirring. The addition of the water reducer can reduce the water demand of the cement paste, thereby improving the strength and durability of the concrete without reducing the workability. Water is a necessary condition for the hydration reaction of cement, and an appropriate amount of water can ensure the full hydration of cement and form a solid concrete structure.

[0056] Accordingly, in the process of adding stone materials and reinforcing fibers to the first cement mixture and stirring to obtain the second cement mixture, the technical concept of this application is to collect the stirring state image of the first cement mixture by a camera, and use image recognition and analysis technology based on computer vision to perform feature analysis and multi-scale perception enhancement of the stirring state image. Based on this, the fusion expression between the stirring state enhancement features of the first scale and the second scale is used to automatically judge whether the stirring state meets the predetermined requirements, and determine whether to add stone materials and reinforcing fibers to the first cement mixture in batches. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of concrete can be ensured, and at the same time, the addition timing of materials can be accurately controlled to avoid premature or late addition, thereby improving the intelligent level of mixture stirring.

[0057] Specifically, in the technical solution of this application, first, the stirring state image of the first cement mixture collected by the camera is obtained. Then, considering that the stirring state image contains the state and distribution of the mixture during the stirring process, etc., these state information are crucial for judging the stirring state. Based on this, in the technical solution of this application, the stirring state image is passed through a stirring state feature extractor based on a dilated convolutional neural network model to capture and refine the key feature information of the halving process, and a stirring state feature map is obtained.

[0058] In an embodiment of this application, passing the stirring state image through a stirring state feature extractor to obtain a stirring state feature map includes: passing the stirring state image through a stirring state feature extractor based on a dilated convolutional neural network model to obtain the stirring state feature map.

[0059] Then, considering that the mixing state feature map contains features of different scales regarding the mixing state, such as color, texture, shape, etc. in the large-scale mixing process at the shallow layer, and high-level semantic features at the small scale in the deep layer, such as the uniformity of mixing, there are also corresponding interferences from some background information. In order to more intensively and meticulously analyze the mixing state feature information of different scales, thereby perceiving the mixing states of different scales during the mixing process, avoiding the omission of small-scale mixing states, and enhancing the multi-layer semantic perception ability of the mixing state, so as to better understand the complexity of the mixing process, in the technical solution of this application, multi-scale perception enhancement expression is performed on the mixing state feature map to obtain the first-scale mixing state perception enhancement feature map and the second-scale mixing state perception enhancement feature map. It should be understood that multi-scale perception enhancement expression enhances the representation ability of mixing states of different scales by performing feature expressions of different scales on the mixing state feature map, thereby highlighting the most important features in the mixing state, such as the uniformity of material mixing or potential separation areas, reducing the interference of background noise, and thus improving the understanding of the overall situation and local details. Specifically, first, convolutional processing based on different scales is performed on the mixing state feature map to obtain shallow-layer and deep-layer features. Then, the deep-layer features are subjected to convolutional encoding processing to further highlight more representative and discriminative mixing semantic features, so as to better capture the essential attributes of the mixing process, and the first-scale mixing state perception enhancement feature map is obtained.

[0060] In an embodiment of this application, performing multi-scale perception enhancement expression on the mixing state feature map to obtain the first-scale mixing state perception enhancement feature map and the second-scale mixing state perception enhancement feature map includes: performing convolutional processing on the mixing state feature map based on multiple convolutional layers to obtain a shallow-layer mixing state feature map and a deep-layer mixing state feature map; performing convolutional encoding on the deep-layer mixing state feature map to obtain the first-scale mixing state perception enhancement feature map; performing upsampling processing on the first-scale mixing state perception enhancement feature map to obtain an upsampled first-scale mixing state perception enhancement feature map; and performing multi-scale feature perception processing on the shallow-layer mixing state feature map to obtain the second-scale mixing state perception enhancement feature map.

[0061] Furthermore, in an embodiment of this application, performing multi-scale feature perception processing on the shallow-layer mixing state feature map to obtain the second-scale mixing state perception enhancement feature map includes: inputting the shallow-layer mixing state feature map into a multi-scale perception network to obtain a shallow-layer multi-scale perception mixing state feature map; fusing the features of the shallow-layer multi-scale perception mixing state feature map and the upsampled first-scale mixing state perception enhancement feature map, and performing convolutional encoding on the obtained mixing state fusion feature map to obtain the second-scale mixing state perception enhancement feature map.

[0062] Next, by inputting the shallow features into the multi-scale perception network, features of different scales in the shallow stirring features are further captured. In particular, the multi-scale perception network first copies the initial shallow features to retain the original stirring state information and uses them as the inputs of two processing branches respectively. Then, one branch directly utilizes the original features, while the other branch successively performs convolutions with 1x1, 3x3, and 5x5 convolutional kernels to progressively learn the features extracted by the previous layer, so as to extract more semantic features. Meanwhile, the output features of each convolutional layer are retained, and activation processing based on the ReLU function and batch normalization are performed after each convolutional layer to refine features of different levels layer by layer, enhancing the expressive ability of the model. After that, the outputs of each convolutional layer and the original features are cascaded to achieve multi-scale perception fusion of features and obtain the shallow multi-scale perception stirring state feature map.

[0063] In an embodiment of the present application, inputting the shallow stirring state feature map into the multi-scale perception network to obtain the shallow multi-scale perception stirring state feature map includes: copying the shallow stirring state feature map to obtain a shallow stirring state copied feature map; first passing the shallow stirring state feature map through a convolutional layer with a 1×1 convolutional kernel, and then successively performing activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a first-scale shallow stirring state feature map; first passing the first-scale shallow stirring state feature map through a convolutional layer with a 3×3 convolutional kernel, and then successively performing the activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a second-scale shallow stirring state feature map; first passing the second-scale shallow stirring state feature map through a convolutional layer with a 5×5 convolutional kernel, and then successively performing the activation processing based on the ReLU function and batch normalization on the obtained feature map to obtain a third-scale shallow stirring state feature map; cascading the shallow stirring state copied feature map, the first-scale shallow stirring state feature map, the second-scale shallow stirring state feature map, and the third-scale shallow stirring state feature map through a feature fusion layer to obtain the shallow multi-scale perception stirring state feature map.

[0064] Furthermore, the first-scale stirring state perception enhanced feature map is upsampled so that feature maps of different scales are consistent in spatial size to obtain an upsampled first-scale stirring state perception enhanced feature map. Finally, the shallow multi-scale perception stirring state feature map and the upsampled first-scale stirring state perception enhanced feature map are feature-fused and then convolutionally encoded to obtain a second-scale stirring state perception enhanced feature map. In this way, by performing multi-scale perception on the stirring state feature map, different aspects of the stirring state can be captured more meticulously and comprehensively, thus contributing to improving the accuracy of stirring state recognition.

[0065] Then, in order to enable the network to focus on more important global information of the stirring state, in the technical solution of this application, global average pooling is performed on each feature matrix along the channel dimension in the first-scale stirring state perception enhancement feature map to obtain a first-scale stirring state perception enhancement feature vector. That is, through global average pooling, high-dimensional features can be compressed into a lower-dimensional feature vector, effectively reducing the feature dimension while retaining important feature information of the stirring state.

[0066] Furthermore, considering that the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map respectively express different levels of stirring states at different scales, and can effectively process and fuse features at different scales to ensure the full utilization of information captured from multiple levels, in the technical solution of this application, the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map are passed through a gated multi-scale stirring state fusion module based on the imitation Transformer structure to obtain a stirring state multi-scale fusion expression feature map. It is worth mentioning that the gated multi-scale stirring state fusion module based on the imitation Transformer structure realizes the effective fusion of features at different scales by imitating the self-attention mechanism of the Transformer structure and combining with a gated unit. Specifically, first, layer normalization is performed on the second-scale stirring state perception enhanced feature map to eliminate the scale differences between different feature maps, so that features from different scales can be compared and fused at the same scale. Then, point convolution encoding and spatial convolution encoding are performed on the normalized features to capture and refine the spatial context feature information of the stirring features, obtaining a second-scale stirring state perception enhanced spatial deepening feature map, and the features are replicated to retain spatial information. Next, the replicated features and spatial features are reshaped to obtain a backup second-scale stirring state perception enhanced spatial deepening feature matrix and a second-scale stirring state perception enhanced spatial deepening feature matrix as the imitation query feature matrix and the imitation key feature matrix. Furthermore, the self-correlation matrix between the imitation query feature matrix and the imitation key feature matrix is calculated to capture the stirring state correlation at different scales between the two, obtaining a second-scale stirring state perception enhanced local self-correlation library matrix. Further, an attention gated unit based on the GELU function is used to adaptively adjust the self-correlation weights of each eigenvalue in the self-correlation matrix to achieve dynamic selection and enhancement of features, obtaining a second-scale stirring state perception enhanced gated local self-correlation library weight matrix. Subsequently, the self-correlation library weight matrix and the first-scale stirring state perception enhanced feature vector are multiplied matrix-wise to achieve feature fusion between different scales, obtaining a stirring state multi-scale cross-modal interaction fusion conditional weight vector. Finally, the conditional weight vector and the second-scale stirring state perception enhanced feature map are weighted channel by channel to obtain a stirring state multi-scale fusion expression feature map that integrates stirring information at different scales.

[0067] In an embodiment of the present application, the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map are passed through a gated multi-scale stirring state fusion module based on an imitation transformer structure to obtain a stirring state multi-scale fusion expression feature map, including: performing layer normalization and convolutional encoding processing on the second-scale stirring state perception enhanced feature map to obtain a second-scale stirring state perception enhanced spatial deepening feature map; copying the second-scale stirring state perception enhanced spatial deepening feature map to obtain a backup second-scale stirring state perception enhanced spatial deepening feature map; respectively performing feature shape reshaping on the backup second-scale stirring state perception enhanced spatial deepening feature map and the second-scale stirring state perception enhanced spatial deepening feature map to obtain a backup second-scale stirring state perception enhanced spatial deepening feature matrix and a second-scale stirring state perception enhanced spatial deepening feature matrix as an imitation query feature matrix and an imitation key feature matrix; calculating the product between the imitation query feature matrix and the transposed matrix of the imitation key feature matrix to obtain a second-scale stirring state perception enhanced local self-correlation library matrix; inputting the second-scale stirring state perception enhanced local self-correlation library matrix into an attention gated unit based on the GELU function to obtain a second-scale stirring state perception enhanced gated local self-correlation library weight matrix; using the first-scale stirring state perception enhanced feature vector as a value feature vector, calculating the product between the second-scale stirring state perception enhanced gated local self-correlation library weight matrix and the first-scale stirring state perception enhanced feature vector to obtain a stirring state multi-scale cross-modal interaction fusion conditional weight vector; using the stirring state multi-scale cross-modal interaction fusion conditional weight vector as a weight vector, performing channel-wise weighting on the second-scale stirring state perception enhanced feature map to obtain the stirring state multi-scale fusion expression feature map.

[0068] Further, in an embodiment of the present application, performing layer normalization and convolutional encoding processing on the second-scale stirring state perception enhanced feature map to obtain a second-scale stirring state perception enhanced spatial deepening feature map includes: performing layer normalization processing on the second-scale stirring state perception enhanced feature map to obtain a layer-normalized second-scale stirring state perception enhanced feature map; performing point convolutional encoding and spatial convolutional encoding on the layer-normalized second-scale stirring state perception enhanced feature map to obtain the second-scale stirring state perception enhanced spatial deepening feature map.

[0069] Specifically, the first-scale stirring state perception enhanced feature vector and the second-scale stirring state perception enhanced feature map are passed through a gated multi-scale stirring state fusion module based on an imitation transformer structure and processed according to the following fusion formula to obtain the stirring state multi-scale fusion expression feature map; where the fusion formula is: ; where is the second-scale stirring state perception enhancement feature map, is a layer normalization operation, is a point convolution encoding, is a convolution operation with a convolution kernel of is the second-scale stirring state perception enhancement spatial deepening feature map, is a copy operation, is the backup second-scale stirring state perception enhancement spatial deepening feature map, is a shape reshaping operation, is the transpose of the feature matrix, is a matrix multiplication, is the GELU function, is the second-scale stirring state perception enhancement gated local self-correlation library weight matrix, is the first-scale stirring state perception enhancement feature vector, is the stirring state multi-scale cross-modal interaction fusion conditional weight vector, is the stirring state multi-scale fusion expression feature map.

[0070] Subsequently, the stirring state multi-scale fusion expression feature map is passed through a stirring state recognizer based on a classifier to obtain a recognition result, and the recognition result is used to indicate whether the stirring state meets a predetermined requirement. In response to the recognition result indicating that the stirring state meets the predetermined requirement, stone materials and reinforcing fibers are added to the first cement mixture in batches. That is, the stirring state multi-scale fusion expression feature map obtained by multi-scale fusion of the first-scale stirring state perception enhancement feature vector and the second-scale stirring state perception enhancement feature map is classified to automatically determine whether the stirring state meets the predetermined requirement and to determine whether to add stone materials and reinforcing fibers to the first cement mixture in batches. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of the concrete can be ensured, and at the same time, the addition timing of the materials can be precisely controlled to avoid premature or late addition, thereby improving the intelligent level of the mixture stirring.

[0071] In an embodiment of the present application, based on the stirring state multi-scale fusion expression feature map, a recognition result is obtained, and based on the recognition result, it is determined whether to add stone materials and reinforcing fibers to the first cement mixture in batches, including: passing the stirring state multi-scale fusion expression feature map through a stirring state recognizer based on a classifier to obtain the recognition result, and the recognition result is used to indicate whether the stirring state meets a predetermined requirement; in response to the recognition result indicating that the stirring state meets the predetermined requirement, adding stone materials and reinforcing fibers to the first cement mixture in batches.

[0072] ​Further, in an embodiment of the present application, obtaining the recognition result by passing the multi-scale fusion expression feature map of the stirring state through a stirring state recognizer based on a classifier includes: expanding the multi-scale fusion expression feature map of the stirring state into a classification feature vector according to a row vector or a column vector; using multiple fully connected layers of the stirring state recognizer based on the classifier to perform fully connected encoding on the classification feature vector to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the Softmax classification function of the stirring state recognizer based on the classifier to obtain the recognition result.

[0073] Preferably, considering that the first-scale stirring state perception enhancement feature map and the second-scale stirring state perception enhancement feature map respectively represent the image semantic features of different-scale perception enhancements of the stirring state image, when performing global mean pooling processing on the first-scale stirring state perception enhancement feature map to obtain a first-scale stirring state perception enhancement feature vector, and passing the first-scale stirring state perception enhancement feature vector and the second-scale stirring state perception enhancement feature map through a gated multi-scale stirring state fusion module based on the transformer structure, there will also be an offset in the aggregation mapping of the image semantic features caused by the inconsistent distribution of the image semantic features under different scales and different distribution dimensions. Therefore, it is desired to further improve the comprehensibility of the fusion of the image semantic features of the multi-scale fusion expression feature map of the stirring state, thereby improving the accuracy of the recognition result obtained by passing the multi-scale fusion expression feature map of the stirring state through a stirring state recognizer based on a classifier.

[0074] Therefore, when passing the multi-scale fusion expression feature map of the stirring state through a stirring state recognizer based on a classifier in the present application, the multi-scale fusion expression feature map of the stirring state is optimized, including the steps of:

[0075] Expanding the multi-scale fusion expression feature map of the stirring state into a multi-scale fusion expression feature vector of the stirring state;

[0076] Subtracting the 0-norm of the multi-scale fusion expression feature vector of the stirring state from the length of the multi-scale fusion expression feature vector of the stirring state to obtain an isolated representation value of the multi-scale fusion expression of the stirring state;

[0077] Calculating the logarithm to the base 2 of the sum of the square of the isolated representation value of the multi-scale fusion expression of the stirring state and the isolated representation value of the multi-scale fusion expression of the stirring state to obtain an information order value of the multi-scale fusion expression of the stirring state;

[0078] Calculate the power function with each eigenvalue of the multi-scale fusion expression feature vector in the stirring state as the base and the difference between the isolated representation value of the multi-scale fusion expression in the stirring state minus one as the exponent, and perform a dot product with the information order value of the multi-scale fusion expression in the stirring state to obtain the multi-scale fusion expression leading vector in the stirring state;

[0079] After performing a dot product of the multi-scale fusion expression feature vector in the stirring state with the difference between the isolated representation value of the multi-scale fusion expression in the stirring state minus one, then perform a dot product with the reciprocal of the isolated representation value of the multi-scale fusion expression in the stirring state to obtain the multi-scale fusion expression field constraint vector in the stirring state;

[0080] Calculate the exponential function with the natural constant as the base and each eigenvalue of the multi-scale fusion expression field constraint vector in the stirring state as the exponent to obtain the multi-scale fusion expression field bias vector in the stirring state; and

[0081] Perform a dot addition of the multi-scale fusion expression leading vector in the stirring state and the multi-scale fusion expression field bias vector in the stirring state to obtain the optimized multi-scale fusion expression feature vector in the stirring state.

[0082] Among them, the multi-scale fusion expression feature vector in the stirring state, for example, denoted as The optimization process is expressed as: ; is to calculate the power function with each eigenvalue of the multi-scale fusion expression feature vector in the stirring state as the base and the difference between the isolated representation value of the multi-scale fusion expression in the stirring state minus one as the exponent, is the multi-scale fusion expression feature vector in the stirring state, is the 0-norm of the vector, is the length of the multi-scale fusion expression feature vector in the stirring state, is the isolated representation value of the multi-scale fusion expression in the stirring state, is to calculate the exponential function with the natural constant as the base and each eigenvalue of the vector as the exponent, the reciprocal of the isolated representation value of the multi-scale fusion expression in the stirring state, is the logarithm value with base 2, is the optimized multi-scale fusion expression feature vector in the stirring state, is the dot addition by position, is the dot product by position.

[0083] Thus, for the high-dimensional feature manifold of the multi-scale fusion expression feature map of the stirring state, a vector field representation with the eigenvalue of the feature set as the aggregation dimension is used. The superimposed value of the vector field of the multi-scale fusion expression feature vector of the stirring state at the isolated zero position is used as the order information to fix the local position of the eigenvalue of its feature set, and a bias for the reversibility of the feature regression distribution field of the multi-scale fusion expression feature vector of the stirring state is added as a reward to achieve the mapping target tracking of the regression distribution of the multi-scale fusion expression feature vector of the stirring state to the eigenvalue position, so that the feature set of the multi-scale fusion expression feature map of the stirring state perceives the mapping migration to the aggregation distribution, thereby improving the understandability of the image semantic feature fusion mapping of the multi-scale fusion expression feature map of the stirring state and improving the accuracy of the recognition result obtained by the stirring state recognizer based on the classifier. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of concrete can be ensured, and at the same time, the addition timing of materials can be accurately controlled to avoid premature or late addition, thus improving the intelligent level of mixing and stirring.

[0084] In summary, adopting the above solution, the stirring state image of the first cement mixture is collected by a camera, and image recognition and analysis techniques based on computer vision are used to perform feature analysis and multi-scale perception enhancement on the stirring state image. Based on the fusion expression between the stirring state enhancement features of the first scale and the second scale, it is automatically determined whether the stirring state meets the predetermined requirements, and it is determined whether to add stone materials and reinforcing fibers to the first cement mixture in batches. In this way, the stirring state can be objectively evaluated, the uniformity difference caused by human factors can be reduced, the uniformity of concrete can be ensured, and at the same time, the addition timing of materials can be accurately controlled to avoid premature or late addition, thus improving the intelligent level of mixing and stirring.

[0085] Figure 2 is a block diagram of a production preparation system for a special concrete shown according to an exemplary embodiment. As Figure 2 shown, the system 200 includes:

[0086] A grinding and pulverizing processing module 201 for grinding and pulverizing tailings and fly ash to obtain ground tailings and ground fly ash;

[0087] A stirring and mixing processing module 202 for stirring and mixing the ground tailings and the ground fly ash to obtain a raw mixture;

[0088] A first cement mixture generating module 203 for adding cement to the raw mixture and stirring to obtain a first cement mixture;

[0089] The second cement mixture generation module 204 is configured to add stones and reinforcing fibers to the first cement mixture and stir them to obtain a second cement mixture;

[0090] The special concrete generation module 205 is configured to uniformly mix a water reducing agent and water, add them to the second cement mixture, and stir to obtain special concrete;

[0091] Among them, the second cement mixture generation module 204 includes:

[0092] The stirring state image acquisition unit 2041 is configured to acquire the stirring state image of the first cement mixture collected by a camera;

[0093] The stirring state feature extraction unit 2042 is configured to pass the stirring state image through a stirring state feature extractor to obtain a stirring state feature map;

[0094] The multi-scale perception enhancement expression unit 2043 is configured to perform multi-scale perception enhancement expression on the stirring state feature map to obtain a first-scale stirring state perception enhancement feature map and a second-scale stirring state perception enhancement feature map;

[0095] The global average pooling processing unit 2044 is configured to perform global average pooling processing on each feature matrix along the channel dimension in the first-scale stirring state perception enhancement feature map to obtain a first-scale stirring state perception enhancement feature vector;

[0096] The gated multi-scale stirring state fusion unit 2045 is configured to pass the first-scale stirring state perception enhancement feature vector and the second-scale stirring state perception enhancement feature map through a gated multi-scale stirring state fusion module based on the transformer structure to obtain a stirring state multi-scale fusion expression feature map;

[0097] The recognition result determination unit 2046 is configured to obtain a recognition result based on the stirring state multi-scale fusion expression feature map, and based on the recognition result, determine whether to add stones and reinforcing fibers to the first cement mixture in batches.

[0098] Next, refer to Figure 3 , which shows a schematic structural diagram of an electronic device 600 suitable for implementing the embodiments of the present application. The terminal devices in the embodiments of the present application may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3The electronic device shown is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present application.

[0099] As Figure 3 shown, the electronic device 600 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 601, which may perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage device 608 into the random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. The input / output (I / O) interface 605 is also connected to the bus 604.

[0100] Generally, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 the electronic device 600 with various devices is shown, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices may be implemented or had alternatively.

[0101] Specifically, according to the embodiments of the present application, the process described above with reference to the flowchart may be implemented as a computer software program. For example, the embodiments of the present application include a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes program codes for performing the method shown in the flowchart. In such an embodiment, the computer program may be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above functions defined in the method of the embodiments of the present application are executed.

[0102] It should be noted that the above computer-readable medium in this application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of a computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, a computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device. And in this application, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0103] In some embodiments, the client and the server can communicate using any currently known or future-developed network protocol such as HTTP (HyperText Transfer Protocol), and can be interconnected with digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0104] The above computer-readable medium can be included in the above electronic device; it can also exist separately without being assembled into the electronic device.

[0105] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The programming languages include, but are not limited to, object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., by connecting through the Internet using an Internet service provider).

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks can occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0107] The modules described in the embodiments of this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation on the module itself in some cases. For example, the test parameter acquisition module can also be described as "the module for acquiring device test parameters corresponding to the target device".

[0108] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, by way of non-limitation, exemplary types of hardware logic components that can be used include: field programmable gate arrays (FPGA), application specific integrated circuits (ASIC), application specific standard products (ASSP), systems on a chip (SOC), complex programmable logic devices (CPLD), and so on.

[0109] In the context of the present application, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0110] Figure 4 is an application scenario diagram of a production preparation method of a special concrete shown according to an exemplary embodiment. As Figure 4 shown, in this application scenario, first, an agitation state image of the first cement mixture collected by a camera is obtained (for example, as Figure 4 illustrated by C); then, the obtained agitation state image of the first cement mixture is input into a server (for example, as Figure 4 illustrated by S) where a production preparation algorithm of the special concrete is deployed, and the server can process the agitation state image of the first cement mixture based on the production preparation algorithm of the special concrete to determine whether to add stones and reinforcing fibers to the first cement mixture in batches.

[0111] The above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosure concept. For example, technical solutions formed by mutually replacing the above features with (but not limited to) technical features having similar functions disclosed in the present application.

[0112] Moreover, although the operations are depicted in a particular order, this should not be construed as requiring that the operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the foregoing discussion, these should not be construed as limitations on the scope of the present application. Certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, the various features that are described in the context of a single embodiment may also be implemented separately or in any suitable sub-combination in multiple embodiments.

[0113] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementation. With regard to the apparatus in the foregoing embodiments, the specific manner in which each module performs operations has been described in detail in the embodiments related to the method and will not be elaborated herein.

Claims

1. A method for producing special concrete, characterized in that: include: Grinding and pulverizing the tailings and fly ash to obtain ground tailings and ground fly ash; The ground tailings and the ground fly ash are mixed and stirred to obtain an original mixed material; Adding cement to the original mixture and stirring to obtain a first cement mixture; adding stone and reinforcing fiber to the first cement mixture and stirring to obtain a second cement mixture; After the water reducing agent and water are evenly mixed, the mixture is added to the second cement mixture, and stirred to obtain special concrete; The step of adding stone and reinforcing fiber to the first cement mixture and stirring the mixture to obtain a second cement mixture comprises: Acquire a mixing state image of the first cement mixture captured by a camera; Passing the stirring state image through a stirring state feature extractor to obtain a stirring state feature map; Performing multi-scale perception enhancement expression on the stirring state feature map to obtain a first-scale stirring state perception enhancement feature map and a second-scale stirring state perception enhancement feature map; Performing global mean pooling processing on each feature matrix along the channel dimension in the first-scale stirring state perception enhancement feature map to obtain a first-scale stirring state perception enhancement feature vector; The first-scale stirring state perception enhancement feature vector and the second-scale stirring state perception enhancement feature map are passed through a gated multi-scale stirring state fusion module based on an imitation converter structure to obtain a stirring state multi-scale fusion expression feature map; Based on the multi-scale fusion expression feature map of the mixing state, obtaining a recognition result, and based on the recognition result, determining whether to add stone and reinforcing fiber to the first cement mixture in batches; The first-scale stirring state perception enhancement feature vector and the second-scale stirring state perception enhancement feature map are fused through a gated multi-scale stirring state fusion module based on an imitation converter structure to obtain a stirring state multi-scale fusion expression feature map, including: performing layer normalization and convolution coding processing on the second-scale stirring state perception enhancement feature map to obtain a second-scale stirring state perception enhancement spatial deepening feature map; Copying the second-scale stirring state perception enhancement spatial deepening feature map to obtain a backup second-scale stirring state perception enhancement spatial deepening feature map; Reshaping the backup second-scale stirring state perception enhanced space deepening feature map and the second-scale stirring state perception enhanced space deepening feature map respectively to obtain a backup second-scale stirring state perception enhanced space deepening feature matrix and a second-scale stirring state perception enhanced space deepening feature matrix as a simulated query feature matrix and a simulated key feature matrix; Calculating the product of the pseudo-query feature matrix and the transposed matrix of the pseudo-key feature matrix to obtain a second-scale stirring state perception enhanced local autocorrelation library matrix; Inputting the second-scale stirring state perception enhanced local autocorrelation library matrix into the attention gating unit based on the GELU function to obtain the second-scale stirring state perception enhanced gated local autocorrelation library weight matrix; Taking the first-scale stirring state perception enhancement feature vector as the value feature vector, calculating the product between the second-scale stirring state perception enhancement gated local autocorrelation library weight matrix and the first-scale stirring state perception enhancement feature vector to obtain a stirring state multi-scale cross-modal interactive fusion condition weight vector; The stirring state multi-scale cross-modal interactive fusion condition weight vector is used as a weight vector, and the second-scale stirring state perception enhancement feature map is weighted channel by channel to obtain the stirring state multi-scale fusion expression feature map.

2. The method for producing special concrete according to claim 1, characterized in that: Passing the stirring state image through a stirring state feature extractor to obtain a stirring state feature map includes: passing the stirring state image through a stirring state feature extractor based on a hole convolutional neural network model to obtain the stirring state feature map.

3. The method for producing special concrete according to claim 2, characterized in that: Performing multi-scale perceptual enhancement expression on the stirring state feature map to obtain a first-scale stirring state perceptual enhancement feature map and a second-scale stirring state perceptual enhancement feature map, including: The stirring state characteristic map is subjected to convolution processing based on a multi-layer convolution layer to obtain a shallow stirring state characteristic map and a deep stirring state characteristic map; Performing convolution encoding on the deep stirring state feature map to obtain the first-scale stirring state perception enhancement feature map; Performing upsampling processing on the first-scale stirring state perception enhancement feature map to obtain an upsampled first-scale stirring state perception enhancement feature map; Multi-scale feature perception processing is performed on the shallow stirring state feature map to obtain the second-scale stirring state perception enhanced feature map.

4. The method for producing special concrete according to claim 3, characterized in that: Performing multi-scale feature perception processing on the shallow stirring state feature map to obtain the second-scale stirring state perception enhanced feature map includes: Inputting the shallow stirring state characteristic map into a multi-scale perception network to obtain a shallow multi-scale perception stirring state characteristic map; The shallow multi-scale perception stirring state feature map and the upsampled first-scale stirring state perception enhanced feature map are feature fused, and the obtained stirring state fusion feature map is convolutionally encoded to obtain the second-scale stirring state perception enhanced feature map.

5. The method for producing special concrete according to claim 4, characterized in that: Inputting the shallow stirring state feature map into a multi-scale perception network to obtain a shallow multi-scale perception stirring state feature map, including: Copying the shallow stirring state characteristic diagram to obtain a shallow stirring state copy characteristic diagram; The shallow stirring state feature map is first passed through a convolution layer with a convolution kernel of 1×1, and then the obtained feature map is sequentially subjected to activation processing based on a ReLU function and standardized batch processing to obtain a first-scale shallow stirring state feature map; The first-scale shallow stirring state feature map is first passed through a convolution layer with a convolution kernel of 3×3, and then the obtained feature map is sequentially subjected to the activation processing based on the ReLU function and the standardized batch processing to obtain a second-scale shallow stirring state feature map; The second-scale shallow stirring state feature map is first passed through a convolution layer with a convolution kernel of 5×5, and then the obtained feature map is sequentially subjected to the activation processing based on the ReLU function and the standardized batch processing to obtain a third-scale shallow stirring state feature map; The shallow stirring state replication feature map, the first-scale shallow stirring state feature map, the second-scale shallow stirring state feature map and the third-scale shallow stirring state feature map are cascaded through a feature fusion layer to obtain the shallow multi-scale perception stirring state feature map.

6. The method for producing special concrete according to claim 5, characterized in that: The second-scale stirring state perception enhancement feature map is subjected to layer normalization and convolution coding processing to obtain a second-scale stirring state perception enhancement spatial deepening feature map, including: performing layer normalization processing on the second-scale stirring state perception enhancement feature map to obtain a layer-normalized second-scale stirring state perception enhancement feature map; Point convolution encoding and spatial convolution encoding are performed on the layer-normalized second-scale stirring state perception enhancement feature map to obtain the second-scale stirring state perception enhancement spatial deepening feature map.

7. The method for producing special concrete according to claim 6, characterized in that: Obtaining a recognition result based on the multi-scale fusion expression feature map of the stirring state, and judging whether to add stone and reinforcing fiber to the first cement mixture in batches based on the recognition result, including: The stirring state multi-scale fusion expression feature map is passed through a stirring state identifier based on a classifier to obtain the recognition result, and the recognition result is used to indicate whether the stirring state meets the predetermined requirements; In response to the identification result that the mixing state meets the predetermined requirement, stone and reinforcing fiber are added to the first cement mixture in batches.

8. The method for producing special concrete according to claim 7, characterized in that: The stirring state multi-scale fusion expression feature map is passed through a stirring state identifier based on a classifier to obtain the recognition result, including: Expanding the stirring state multi-scale fusion expression feature map into a classification feature vector according to a row vector or a column vector; Performing fully connected encoding on the classification feature vector using a plurality of fully connected layers of the classifier-based stirring state identifier to obtain an encoded classification feature vector; The encoded classification feature vector is passed through the Softmax classification function of the classifier-based stirring state identifier to obtain the identification result.

Images